Can patient-reported outcome measures predict mortality in neurological populations? A systematic review

Patient-reported outcome measures (PROMs) are increasingly used for symptom monitoring and care delivery, yet their prognostic value for identifying patients at higher risk for mortality in neurological populations is unclear. This systematic review evaluated whether PROMs predict mortality and/or survival in adults with neurological conditions.

We systematically searched MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials (January 2002-November 2024) for studies incorporating PROMs into mortality or survival prediction models across 10 neurological conditions: motor neuron disease, diabetic neuropathy, nervous system cancers, Alzheimer's and other dementias, Guillain-Barré syndrome, epilepsy, headache, multiple sclerosis, Parkinson's disease, and stroke. Screening, data extraction, and risk-of-bias assessment followed the CHARMS and PRISMA guidelines. Findings were descriptively summarized.

Of 6,218 abstracts reviewed, 49 studies met the inclusion criteria. Most evaluated stroke (n = 16), nervous system cancers (n = 14), or motor neuron disease (n = 9). None evaluated headache, diabetic neuropathy, Guillain-Barré syndrome, or epilepsy. Of the included studies, 26 used generic PROMs, 19 used condition-specific PROMs, and 4 included both. Across conditions, PROMs independently predicted mortality in three-quarters of studies, with the strongest evidence observed in nervous system cancers and motor neuron disease. By instruments, EORTC QLQ in brain cancers and SF-36 in stroke showed the most consistent prognostic utility. Among studies with mixed findings by domain, physical health components were more likely to predict mortality than emotional components.

PROMs independently predict mortality in several neurological conditions, though prognostic value varied by condition and instrument type. Future studies should evaluate their additive value and feasibility for integration into prognostic models in routine care.